Scuba

Scuba is building the world's largest voice data infrastructure for languages non-existent on the internet. We collect, label, and license high-quality speech corpora for the world's most underrepresented languages, starting with Ibibio & Efik.

Our corpus has 112+ hours of accepted audio across 513 unique speakers, with emotion-labeled prompts, human review by native speakers, and full consent documentation.

For licensing, research partnerships, or dataset access, visit scubavoice.com or email info@scubavoice.com.